Your zZuPer
is not alive.

You are talking to an AI

Every zZuPer — in zZuPer Talk, in a roaming note, in a Pulse icebreaker — is generated by a large language model. It has a personality because we wrote one for it, not because it has one.

We keep the pet in character: it speaks in its species' sounds and does not interrupt itself to remind you it is a program. That is a decision about tone. It is not permission to be deceived about what it is. The header of every zZuPer Talk conversation reads AI companion. The app states it again at the start of a conversation and at least every three hours of continued use. We state it here, and in the Terms:

zZuPer responses are artificially generated. They are not written by a human, and your pet is not a person.

The one place the pet will always drop the act is a crisis. See below.

What your pet does if you are in crisis

We publish this protocol in full because you deserve to know what a piece of software will do at the worst moment, and because the law requires operators of companion chatbots to publish it.

What triggers it

The model is instructed to treat any expression of self-harm, suicidal thinking, or hopelessness about being alive as a crisis — including indirect phrasing like "I don't want to be here anymore".

What it is instructed to do

• Break its own rules. The pet normally answers in one or two short sentences. In a crisis that limit is explicitly lifted, so a length constraint can never crowd out a helpline number.
• Stay warm and stay present as your pet, rather than switching to a cold automated notice.
• Give real, specific resources — not "seek help":
  • 988 Suicide & Crisis Lifeline — call or text 988 (24/7, free, US & Canada)
  • Crisis Text Line — text HOME to 741741
  • Find a Helpline — findahelpline.com, for numbers in your own country
  • Emergency services — call 911, or go to your university's counselling centre right away
• Tell you plainly that you matter, and encourage you to reach a person you trust or a professional now.

What it cannot do — please read this part

• It is an instruction to a model, not a guarantee. Language models do not follow rules perfectly. It can miss phrasing it has not seen, and it can get things wrong.
• Nobody is watching. Pet conversations are not monitored by a human in real time. Telling your pet you are in danger does not alert us, your university, or anyone else.
• It cannot call anyone for you. It cannot dial, text, locate you, or contact your family.

If you are in immediate danger, close the app and call your local emergency number. A cartoon cat is not the right tool, and we would rather say so than let you find out the hard way.

Reporting someone

Both of a person's identities belong to one account, and both are covered by our Community Guidelines. Harassment from a pet is harassment.

Report from inside the app. Three ways:

• Hold any message and tap Report. The report is tied to that exact message.
• Open a profile — a person's or a pet's — and tap Report.
• Profile → Report a problem for anything else.

You can report an anonymous pet without knowing who is behind it. You see a pet and a tag that exists only inside that one conversation. Our servers resolve it. You never see the account, and you do not need to.

You can also report the AI. If your zZuPer says something harmful, report the reply. Nobody is accused — that is not a person, and those reports go to fixing the model.

Screenshots are not required. Reporting inside a conversation captures the message and the surrounding history on our side. Attach images if you want to.

Reports are confidential. We do not tell the person who was reported who reported them. If you cannot get into the app, email admin@zzup.org.

Because messages on zZuP! can never be deleted, the evidence is still there when you report it. That is the entire reason for that design.

What happens next

• We review every report we receive. We aim to review a report within seven days, and sooner where it indicates an immediate risk of harm.
• Sexual content involving minors, credible threats of violence, and encouraging self-harm are acted on immediately, and reported to the authorities where the law requires it.
• We decide what action to take, up to and including permanent removal of the account, at our discretion. We are not required to follow any particular sequence of steps before doing so.
• If we act against your account you can appeal to admin@zzup.org.
• Reporting someone falsely in order to harass them is itself a violation.

Blocking, and why it works per identity

You can block a person's pet without blocking the person, or block them outright. Blocking someone's real identity also ends any friendship between you. Blocking a pet leaves your human-level relationship untouched — and the other person is not told which of their identities you blocked.

Deleting an account does not clear blocks other people placed on it. Someone cannot delete and re-register to get around your block.

Meeting people from zZuP!

The whole point of this app is that strangers become people you know. That is also the part with real-world risk. If you arrange to meet someone: pick somewhere public, tell a friend where you are going and when, and leave whenever you want to. You never owe anyone an explanation for going home.

We do not verify that people are who they say they are. A verified student email means an email was verified — nothing else.

Contact

Safety concerns, reports, appeals and press: admin@zzup.org. A human reads that inbox.